Coronary artery disease (CAD) is the leading cause of death in the United States, even though significant efforts have been made in prevention and diagnosis. The clinical gold standard for diagnosis of CAD is catheter-based invasive x-ray angiography, performed more than a million times per year. Of these examinations, up to 35% have been found to have no significant stenosis, yet these patients had to go through the potential risks and complications of an invasive test that further exposes the patient to ionizing radiation and iodinated contrast. Thus, non-invasive diagnostic alternatives are highly desirable. Cardiac MRI (CMR) provides a method for comprehensive non-invasive cardiac exam, including contractile functional assessment (cine CMR) to detect wall-motion abnormality, myocardial CMR perfusion for diagnosing perfusion defects, viability assessment using late gadolinium enhancement (LGE) for evaluation of acute and chronic myocardial infarction, and coronary MRI for the identification of stenosis. CMR is advantageous in several respects, since it does not require ionizing radiation or iodinated contrast, thereby facilitating repeated or follow-up scanning. However, long data acquisition time remains as one of its main limitations. Several approaches have been studied to facilitate rapid CMR acquisition. Nonetheless, the acquisition time for high-resolution CMR remains long, and spatial and temporal resolution is traded off for acquisition time in cine and perfusion imaging. Therefore, developments of methods to reduce data acquisition time beyond what is available now are appealing. We will develop novel reconstruction methodologies for high-resolution CMR that learn the anatomical structures in the images being reconstructed. We will validate these techniques in a range of contrast-enhanced CMR imaging protocols, providing better volumetric coverage of the heart, efficient use of the contrast agents, and higher spatial and temporal resolution.